Tuesday, September 16, 2008

23 and Me

A while back I was thinking about personal genome sequencing and wondering to myself why, if the entire human genome has been sequenced, do you need to sequence the whole thing again every time? Instead why not just focus on the deltas, the 1% of the genome that differ between people? This is the part that makes us unique and lets you know what diseases you are more susceptible to and how you will react to different drugs. If possible, it seemed like it would be a lot cheaper to do than a complete sequencing.

Then, bam, I read about the company 23 and Me doing just that. I have been meaning to blog on them for a while and now that they have have released version 2.0 of their system with over 650,000 measurements and dropped their price to $399, it is time. The era of affordable personal genomics is now upon us.

The Economist gives a nice overview of the system:

The firm will analyse DNA derived from a saliva sample and post the results on its secure website, which will feature social-networking features (such as finding strangers with similar genetic variations) akin to those of Facebook.

Two keys have unlocked this market: a clever short-cut and a clever chip. All people are 99.9% identical, genetically speaking, and have almost all of the 3 billion letters in the human genetic sequence in common. In a large fraction of the population, only at certain places is one letter replaced by another. These points are called single nucleotide polymorphisms, or SNPs, and there are 10m or so of them. Looking only for SNPs is cheaper and quicker than analysing the entire genome.

The clever chip responsible for that analysis reduces an entire genetics lab to Lilliputian proportions. 23andMe, for example, outsources its genetic analysis to Illumina, whose BeadChip is a glass slide with millions of small beads on its surface. Attached to each of those beads are strands of DNA that are complementary to places in the genome where interesting SNPs are located. When an individual's DNA sample is sliced up and spread over the slide, the fragments are attracted to complementary bits of DNA and stick to them.
The downside to this system is that it doesn't pick up on indels, copy number variations, inversions or so called "junk" DNA that produces around 1/2 a million varieties of RNA of unknown functions. But for just $399 it beats the heck out of the $10 million it cost Craig Venter to sequence his diploid genome or the $350,000 which Knome charges to do whole-genome sequencing.

There are currently 89 conditions for which 23 and Me can tell you about (you can sign up for a demo account on their site to check them out). Of those only 29 are based on what they call "established research". But, as more research is done, the number will grow. They have tests for serious diseases like prostate cancer, diabetes, and multiple sclerosis as well as lighter material like earwax type, odor detection, and freckling.

While some worry about privacy, whether taking the test will allow insurance companies to deny you coverage and other Gattaca related issues, others are more than willing to share their results. Andrew Meyer of Buzzyeah got people to donate $1000 to pay the fee by promising to put the results online. Amy Harmon of the NY Times and Michael Arrington of TechCrunch both wrote about their results. Mike Spear went the extra mile and is giving access to his results both in 23 and Me and in Promethease which annotates your genome using SNPedia.

If I were 23 and Me, I would market this (or a stripped down cheaper version) to parents when their children enter school. I think it would be useful for parents to know how their child's genome will impact them. In particular I would want to know how they will react to drugs. 23 and Me tests for alcoholism, the alcohol flush face reaction, nicotine dependence, and slow or fast caffeine metabolism (but surprisingly not for cocaine addiction). Knowing that my child is likely to get hooked on alcohol or cigarettes would make me more vigilant about making sure they never start.

I am right on the fence as to whether I sign up now or if I wait a year or two for the price to come down further and have more conditions tested for. Going through the list, I didn't see any test that would be actionable information in the short run. But, it would be cool to join the personal genomics revolution, and $400 isn't too steep.

Update: Forgot to mention this in depth Wired article on 23 and Me.

3 comments:

cariaso said...

If you've only seen Mike Spear's promethease report, then you're a bit out of date. This is one of the newer reports

http://www.snpedia.com/files/promethease/outputs/promethease-lilly-mendel.html

which show medicines and diseases as distinct categories.

Fat Knowledge said...

Thanks for the pointer Cariaso.

Anonymous said...

As of early 2010, it does pick up indels, at least in some cases. I didn't think it did, but I am a known carrier for a connexin 26 mutation (deafness) that is a single base deletion, and 23andMe's analysis was able to detect it.

Post a Comment

Note: Only a member of this blog may post a comment.